A basic optical signal channel comprises an optical information channel, which for a guided channel, can generally be referred to as a waveguide. The waveguide conducts an optical signal from a first optical component to a second optical component. The optical signal may be light of a single frequency (or color), or may be a multiplexed combination of optical signals of different frequencies in a wavelength multiplexing scheme.
When the waveguide is formed as part of an integrated circuit on a microchip for transmission of the optical signal between components, the on-chip waveguide may be formed as a free-space optical path, or more commonly, a photolithographically produced waveguide core material that is surrounded by boundary materials, or cladding. A boundary material has an index of refraction (IOR) that is different from (generally lower) than the IOR of the core. With appropriate selections of IORs for the core and boundary materials based on the frequency characteristics of a particular optical signal, the signal is transmitted through the waveguide core material.
A typical cross section of a basic on-chip waveguide 10 is illustrated in FIG. 1. Core material 12 is surrounded by one or more types of boundary materials which, in this example, are identified as bottom boundary material 14, top boundary material 16 and side boundary material 18. Core material 12 is the optical waveguide material through which a light signal is transmitted. The difference between the IOR of core material 12 and the boundary materials can be as small as 0.001 for a functional optical waveguide. Further the state of the top and bottom boundary materials is not limited. A boundary material may be in the gaseous, liquid or solid state, as long as the material satisfies the differential IOR relationship between boundary materials and the core to permit optical waveguiding. While it may be advantageous for light propagation symmetry to have the IOR of the bottom and top boundary materials equal to the IOR of the side boundary material, it is not necessary, as long as the IOR of all boundary materials is lower than the IOR of core material 12. FIG. 1 also shows a substrate 20 upon which bottom boundary layer 14 is disposed. In other applications the substrate itself may be the bottom boundary layer. For further reference to conventional on-chip optical waveguides see the Handbook of Photonics, Gupta, Mool C., editor-in-chief, CRC Press LLC, Boca Raton, Fla., 1997.
Conventional on-chip optical waveguides are formed from either organic or inorganic materials using conventional integrated circuit fabrication and patterning techniques. While these materials, such as silicon dioxide and quartz, are similar to those used in a fiber optic cable, light signal transmission losses through an on-chip waveguide is considerably greater than those experienced through an optical fiber. Light wave propagation losses in an optical waveguide are typically from two sources. The first is optical absorption, or scattering, in the bulk of the waveguide material, while the second is interface scattering from the light interaction with the walls of the waveguide. The conventional fabrication technique for the waveguide core, which requires a blanket layer deposition of a material and subsequent selective removal of the material by photoresist patterning and wet, or dry, chemical etching, results in wall damage of the core that increases the core-boundary material interface scattering losses.
In a conventional process for forming waveguide 10, as illustrated in FIG. 2(a) through FIG. 2(f), waveguide base material 22 is blanket deposited on bottom boundary material 24, which in this example, also serves as the substrate. Photoresist 26 is deposited on waveguide base material 22 (FIG. 2(a)). Photoresist 26 is typically a spun-on organic material that completes crosslinking with selective exposure to ultraviolet (UV) light through a mask 60, and subsequent baking (FIG. 2(b)). Photoresist 26 is selectively developed to leave a photoresist mask 61 over the desired waveguide core, while the remainder of photoresist 26 is etched away by a conventional etching method (FIG. 2 (c)). Excess waveguide base material 22 is etched away by conventional etching methods to form waveguide core 22a, with side walls 34, under photoresist mask 61 (FIG. 2(d)). Then the photoresist mask is etched away (FIG. 2(e)). To complete fabrication of waveguide 10, suitable side boundary material 30 and upper boundary material 32 are deposited around waveguide core 22a (FIG. 2(f)). The etching process that removed the excess waveguide base material 22 resulted in irregularities in the side walls 34 of the waveguide core that increase interface scattering losses for light signals transmitted in the waveguide core.
Also known are optical waveguides formed from material diffusion processes. For example, titanium may be selectively diffused into regions of lithium niobate to form an optical waveguide wherein the diffused regions have a higher IOR than the non-diffused regions.
Therefore there is the need for an on-chip optical waveguide that can be fabricated without the boundaries of the waveguide core being subjected to photoresist etch damage and without the diffusion of a material into the base waveguide material.
With respect to organosilicons that might serve as plasma deposited waveguide material, despite intensive research on the plasma deposition of amorphous silicon from monosilane (SiH4), there have been only a few reports exploring the formation of Si—Si bonded polymers from monosubstituted organosilanes. Haller reported an example of selective dehydrogenative polymerization, but no photochemical studies were described. See Hailer, Journal of the Electrochemical Society A, Vol. 129, 1987, p. 180, and Inagaki and Hirao, Journal of Polymer Science A, Vol. 24, 1986, p. 595. Studies on the plasma chemistry of methylsilane (MeSiH3) have involved higher radio-frequency powers and temperatures which promote formation of amorphous silicon carbide (SiC) rather than reactive polymeric product. See Delpancke, Powers, Vandertop and Somorjai, Thin Solid Films, Vol. 202, 1991, p. 289. Low power plasma polymerization of tetramethylsilane and related precursors has been proposed to result in the formation of Si—C—Si linkages. See Wrobel and Wertheimer, Plasma Deposition, Treatment and Etching of Polymers, Academic Press, New York, Chapter 3. Such materials lack sufficient absorption in light above approximately 225 nm wavelength, but have been studied as far ultraviolet (193 nm wavelength) resists by Horn and associates. See Horn, Pang and Rothschild, Journal of Vacuum Science Technology B, Vol. 8, 1991, p. 1493. Polymer chemistry teaches the use of the basic silanes are insignificant as a monomer for polymerization type of polymer. Furthermore, polysiloxanes are differentiated from the basic silanes, and contrasted as being very important in terms of monomers for polymerization. See Stevens, Malcom P., Polymer Chemistry, An Introduction, Addison-Wesley Publishing Co., 1975: p. 334.
Work has been reported on the synthesis of soluble poly-alkylsilyne network polymers ([SiR]n) which exhibit intense ultraviolet absorption (associated with extended Si—Si bonding) and may be photo-oxidatively patterned to give stable siloxane networks. See Bianconi and Weidman, Journal of the American Chemical Society, Vol. 110, 1988, p. 2341. Dry development is accomplished by selective anisotropic removal of unexposed material by chlorine or hydrobromic acid reactive ion etching. See Hornak, Weidman and Kwock, Journal of Applied Physics, Vol. 67,1990, p. 2235, and Horn, Pang and Rothschild, Journal of Vacuum Science Technology B, Vol. 8, 1991, p. 1493. The exposed, oxidized material may be removed by either wet or dry fluorine based chemistry. Kunz and associates have shown that this makes polysilynes particularly effective as 193 nm wavelength photoresists. See Kunz, Bianconi, Horn, Paladugu, Shaver, Smith, and Freed, Proceedings of the Society of Photo-optical and Instrumentation Engineers, Vol. 218, 1991, p. 1466. The high absorbability and the wavelength limits photo-oxidation to the surface, eliminating reflection, and the pattern is transferred through the remainder of the film during the reactive ion etch (RIE) development. Studies of organosilicon hydride network materials containing reactive R—Si—H moieties have found that such high silicon compositions as [MeSiH0.5]n exhibit superior photosensitivity and function as single layer photodefinable glass etch masks. See Weidman and Joshi, New Photodefinable Glass Etch Masks for Entirely Dry Photolithography: Plasma Deposited Organosilicon Hydride Polymers, Applied Physics Letters, Vol. 62, No. 4, 1993, p. 372. However, cost and availability of the exotic organosilicon feedstocks have significantly inhibited the transfer of such photosensitive organosilicon hydride network materials into microcircuit fabrication. Further, films deposited from single component organosilicon feedstocks possess limited latitude in alteration of deposited film characteristics, such as the radiation frequency of photosensitivity and selectivity during etch processes.